mirror of
https://github.com/TangentFoxy/lua-htmlparser.git
synced 2025-07-28 11:02:18 +00:00
improved not and lost method chaining
fixed :not() in that it filters after all matches, preventing later selection of elements that shouldn't have been there Also, ditched the idea of returning a container node, since it was complex and didn't add much. The functionality could be reintroduced by having Set implement the __call or maybe even __index to return the combined results of all its elements.
This commit is contained in:
@@ -17,9 +17,7 @@ function ElementNode:new(nameortext, node, descend, openstart, openend)
|
||||
deepernodes = Set:new(),
|
||||
deeperelements = {}, deeperattributes = {}, deeperids = {}, deeperclasses = {}
|
||||
}
|
||||
if nameortext == "container" then
|
||||
instance.root = node
|
||||
elseif not node then
|
||||
if not node then
|
||||
instance.name = "root"
|
||||
instance.root = instance
|
||||
instance._text = nameortext
|
||||
@@ -90,58 +88,41 @@ function ElementNode:close(closestart, closeend)
|
||||
end
|
||||
|
||||
local function select(self, s)
|
||||
if not s or type(s) ~= "string" then return {} end
|
||||
local subjects = Set:new({self})
|
||||
local resultset
|
||||
local childrenonly
|
||||
if not s or type(s) ~= "string" then return Set:new() end
|
||||
local subjects, resultset, childrenonly = Set:new({self})
|
||||
local sets = {
|
||||
[""] = self.deeperelements,
|
||||
["["] = self.deeperattributes,
|
||||
["#"] = self.deeperids,
|
||||
["."] = self.deeperclasses
|
||||
}
|
||||
for part in string.gmatch(s, "%S+") do
|
||||
if part == ">" then childrenonly = true goto nextpart end
|
||||
resultset = Set:new()
|
||||
for subject in pairs(subjects) do
|
||||
local init = subject.deepernodes
|
||||
if childrenonly then init = Set:new(subject.nodes) childrenonly = false end
|
||||
resultset = resultset + init
|
||||
local star = subject.deepernodes
|
||||
if childrenonly then star = Set:new(subject.nodes) end
|
||||
childrenonly = false
|
||||
resultset = resultset + star
|
||||
end
|
||||
if part == "*" then goto nextpart end
|
||||
local match, filter
|
||||
local excludes, filter = Set:new()
|
||||
for t, w in string.gmatch(part, "([:%[#.]?)([^:%(%[#.%]%)]+)%]?%)?") do
|
||||
-- TODO tidy up
|
||||
if t == ":" then filter = w goto nextw end
|
||||
if t == "" then match = self.deeperelements[w]
|
||||
elseif t == "[" then match = self.deeperattributes[w]
|
||||
elseif t == "#" then match = self.deeperids[w]
|
||||
elseif t == "." then match = self.deeperclasses[w]
|
||||
end
|
||||
local match = sets[t][w]
|
||||
if filter == "not" then
|
||||
resultset = resultset - match
|
||||
excludes = excludes + match
|
||||
else
|
||||
resultset = resultset * match
|
||||
end
|
||||
filter = nil
|
||||
::nextw::
|
||||
end
|
||||
resultset = resultset - excludes
|
||||
subjects = Set:new(resultset)
|
||||
::nextpart::
|
||||
end
|
||||
-- construct a container node for the resultset, so that we can :select() on it
|
||||
local ret = ElementNode:new("container", self)
|
||||
for node in pairs(resultset) do
|
||||
table.insert(ret.nodes, node)
|
||||
ret.deepernodes = ret.deepernodes + node.deepernodes
|
||||
for listname,list in pairs({
|
||||
deeperelements = node.deeperelements,
|
||||
deeperattributes = node.deeperattributes,
|
||||
deeperids = node.deeperids,
|
||||
deeperclasses = node.deeperclasses
|
||||
}) do
|
||||
local target = ret[listname]
|
||||
for k,set in pairs(list) do
|
||||
-- Set.__add will create an empty Set if not target[k]
|
||||
target[k] = target[k] + set
|
||||
end
|
||||
end
|
||||
end
|
||||
return ret
|
||||
return resultset
|
||||
end
|
||||
|
||||
function ElementNode:select(s) return select(self, s) end
|
||||
|
Reference in New Issue
Block a user